Professional Documents
Culture Documents
Mise en pratique
for the definition of the kilogram
in the SI
Consultative Committee for Mass and Related Quantities
1. Introduction
The purpose of this mise en pratique, prepared by the Consultative Committee for Mass and
Related Quantities (CCM) of the International Committee for Weights and Measures (CIPM), is
to indicate how the definition of the SI base unit, the kilogram, symbol kg, may be realized in
practice.
In general, the term “to realize a unit” is interpreted to mean the establishment of the value and
associated uncertainty of a quantity of the same kind as the unit that is consistent with the
definition of the unit. The definition of the kilogram does not imply any particular experiment for
its practical realization. Any method capable of deriving a mass value traceable to the set of
seven reference constants could, in principle, be used. Thus, the list of methods given is not
meant to be an exhaustive list of all possibilities, but rather a list of those methods that are easiest
to implement and/or that provide the smallest uncertainties and which are officially recognized as
primary methods by the relevant Consultative Committee.
A primary method is a method having the highest metrological properties; whose operation can
be completely described and understood; for which a complete uncertainty statement can be
written down in terms of SI units; and which does not require a reference standard of the same
quantity.
2.1. Definition
The definition of the kilogram, SI base unit of mass, is as follows [2.1]:
The kilogram, symbol kg, is the SI unit of mass. It is defined by taking the fixed
numerical value of the Planck constant h to be 6.626 070 15 × 10–34 when expressed
in the unit J s, which is equal to kg m2 s–1, where the metre and the second are
defined in terms of c and ∆νCs.
Thus the Planck constant h is exactly h = 6.626 070 15 × 10–34 J s. This numerical value of h
defines the unit joule second in the SI and, in combination with the SI second and metre, defines
the kilogram. The second and metre are themselves defined by exact values of the hyperfine
transition frequency ∆νCs of the caesium 133 atom and the speed of light in vacuum c. The
numerical value of h given in the definition of the kilogram has ensured the continuity of the unit
of mass with the previous definition, as explained in section 5.
1/14
SI Brochure – 9th edition (2019) – Appendix 2 20 May 2019
The current primary methods focus on the realization and dissemination of the unit of mass at a
nominal value of 1 kg. The mise en pratique may be updated to include information on primary
methods at different nominal mass values.
Primary methods for the realization of the definition of the kilogram and procedures for its
dissemination through primary mass standards are described in the following two sections. The
traceability chain is shown schematically in Figure 1.
Figure 1. Illustration of the traceability chain from the definition of the kilogram. The unit of the Planck
constant being kg m2 s-1, the units second and metre are needed to derive a primary mass standard from the
Planck constant.
This mise en pratique will be updated to take account of new methods and technological
improvements. It is not printed in the SI Brochure [2.1], but the current version is posted on the
open BIPM web site at http://www.bipm.org/en/publications/si-brochure/.
2/14
SI Brochure – 9th edition (2019) – Appendix 2 20 May 2019
There are currently two independent primary methods that are capable of realizing the definition
of the kilogram with relative uncertainties within a few parts in 108. The first of these relies on
determining the unknown mass using an electromechanical balance specially designed for the
purpose. The second method compares the unknown mass to the mass of a single atom of a
specified isotope by counting the number of atoms in a crystal, where the mass of the atom is
well-known in terms of h, c and ∆νCs.
The determination of the unknown mass mx of an artefact x is carried out in two modes: the
weighing mode and the moving mode. They may occur successively or simultaneously. In the
weighing mode, the weight 2 mx g of the artefact is balanced by the electromagnetic force
produced, for example, on a circular coil of wire-length l immersed in a radial magnetic field of
flux density B when a current I1 flows through the coil. The magnet and coil geometries are
designed to produce a force that is parallel to the local gravitational acceleration. The acceleration
of gravity g acting on the mass, and the current I1 flowing in the coil are measured simultaneously
so that
mx g = I1 B l. (3.1)
In the moving mode, the voltage U2 which is induced across the terminals of the same coil
moving vertically at a velocity v through the same magnetic flux density, is measured so that
U2 = v B l. (3.2)
The equations describing the two modes are combined by eliminating Bl:
mx g v = I1 U2 . (3.3)
Thus power of a mechanical nature is equated to power of an electromagnetic nature. The powers
are manifestly “virtual” in this method of operation because power does not figure in either mode
of this two-mode experiment.
The current I1 can, for example, be determined using Ohm’s law by measuring the voltage drop
U1 across the terminals of a stable resistor of value R. Both voltages, U1 and U2, are measured in
terms of the Josephson constant KJ, which is taken to be KJ = 2e/h; e is the elementary charge.
Similarly, R is measured in terms of the von Klitzing constant RK, which is taken to be RK = h/e2.
The quantities v and g are measured in their respective SI units, m s-1 and m s-2.
Note that KJ2RK = 4/h allowing (3.3) to be rewritten schematically as
1
We refer to watt balances as “Kibble balances” to recognize Dr. Bryan Kibble, who originally conceived
the idea of this experiment.
2
In legal metrology “weight” can refer to a material object or to a gravitational force. The terms “weight
force” and “weight piece” are used in legal metrology if the meaning of “weight” is not clear from the
context [3.1].
3/14
SI Brochure – 9th edition (2019) – Appendix 2 20 May 2019
bf 2 1
mx = h , (3.4)
4 gv
All relevant influences on the mass, mx, as derived from (3.4) must be considered for the
realization, maintenance and dissemination of the unit of mass (see also Annex A2).
Other electromagnetic and electrostatic realizations have been proposed, such as the joule-
balance and volt-balance methods, and may well be perfected [3.2, 3.3].
N = 8Vs/a(28Si)3, (3.5)
where 8 is the number of atoms per unit cell of crystalline silicon and a(28Si)3 is the volume of the
unit cell, which is a cube; i.e., Vs/a(28Si)3 is the number of unit cells in the crystal and each unit
cell contains eight silicon 28 atoms. Since the volume of any solid is a function of temperature
and, to a lesser extent, hydrostatic pressure, Vs and a(28Si)3 are referred to the same reference
conditions. For practical reasons, the crystal is fashioned into a sphere having a mass of
approximately 1 kg.
To realize the definition of the kilogram, the mass ms of the sphere is first expressed in terms of
the mass of a single atom, using the XRCD method 4:
ms = N m(28Si), (3.6)
Since the experimental value of the physical constant h/m(28Si) is known to high accuracy [3.5],
one can rewrite (3.6) as
3
The measurements described here were first used to determine the value of the Avogadro constant NA,
which is defined as the number of elementary entities per mole of substance. An accurate measurement of
NA was an essential contribution on the road to redefining the kilogram in 2018. Today, however, the
numerical value of NA is exactly defined when expressed in the SI unit mol-1 thus making the definition of
the mole independent of the kilogram.
4
It is well known that (3.6) is not exact because the right-hand side is reduced by the mass equivalent, E/c2,
of the total binding energy E of the atoms in the crystal, where c is the speed of light in vacuum. The
correction, about 2 parts in 1010 [3.4], is insignificant compared with present experimental uncertainties and
has therefore been ignored. Additional energy terms (e.g. thermal energy) are even smaller than the binding
energy and thus negligible.
4/14
SI Brochure – 9th edition (2019) – Appendix 2 20 May 2019
m( 28 Si)
ms = h N . (3.7)
h
The XRCD experiment determines N; m(28Si)/h is a constant of nature whose value is known to
high accuracy and, of course, the numerical value of h is now fixed.
The sphere is a primary mass standard and the unit of mass, the kilogram, is disseminated from
this standard. Spheres currently used in this work are enriched in the isotope 28Si but the presence
of trace amounts of two additional silicon isotopes leads to obvious modifications of the simple
equations presented in this section. See [3.6] for a more complete analysis of this experiment.
All relevant influences on the mass of the sphere, ms, as derived from (3.7) must be considered
for the realization, maintenance and dissemination of the unit of mass (see also Annex A2).
The definition of the kilogram ensures that the unit of mass is constant in time and that the
definition can be realized by any laboratory, or collaboration of laboratories, with the means to do
so. Any National Metrology Institute (NMI), Designated Institute (DI), the Bureau International
des Poids et Mesures (BIPM), or collaboration among them, that realizes the kilogram definition
can disseminate the SI kilogram from its primary mass standards to any other laboratory or, more
generally, to any user of secondary mass standards (see Figure 1). This is described in section
4.1 5. Dissemination from a dedicated ensemble of 1 kg secondary standards maintained at the
BIPM, called BIPM ensemble of reference mass standards, is described in section 4.2.
The BIPM in coordination with the CCM organizes an on-going BIPM key comparison [4.5],
BIPM.M-K1 [4.6], for laboratories with primary realization methods. In this comparison, the
primary mass standards of the participants are compared to artefacts from the BIPM ensemble of
reference mass standards (see section 4.2). The CCM decides the required periodicity of
laboratory participation in BIPM.M-K1 in order to support relevant calibration and measurement
capabilities (CMCs).
In cases where compliance with the CIPM Mutual Recognition Arrangement (CIPM MRA) is
required [4.7], it is essential that the mass standards are traceable to primary mass standards of a
participant in BIPM.M-K1 that has relevant CMC entries or, in the case of the BIPM, suitable
entries in its calibration and measurement services as approved by the CIPM. Dissemination of
5
In order to preserve the international equivalence of calibration certificates, the National Metrology
Institutes having a realization of the kilogram avail themselves of the consensus value (output of a
statistical analysis of all the data from available realizations of the kilogram to be used as the highest
source of traceability to the redefined kilogram before the dissemination of individual realizations. The
consensus value is managed by a CCM task group to ensure stability and continuity, taking all new
realizations and comparisons into account. It could be identical to the Key Comparison Reference Value
(KCRV) but could also be calculated using additional weighting factors) when disseminating the unit of
mass until the dispersion of the results from individual realization experiments is compatible with the
uncertainties of the individual realizations [4.1]. See also [4.2], [4.3], [4.4] and Annex A3, which all
address issues related to the dissemination of the kilogram from multiple realizations of its definition.
5/14
SI Brochure – 9th edition (2019) – Appendix 2 20 May 2019
the whole mass scale is validated for all NMIs/DIs and the BIPM through the traditional types of
key comparisons organized prior to the present definition of the kilogram.
Results of all key comparisons are published in the Key Comparison Database (KCDB) in
accordance with the rules of CIPM MRA [4.5] and may be used in support of NMI/DI claims of
its calibration and measurement capabilities and the BIPM claims listed in its calibration and
measurement services.
The average mass of the ensemble is derived from links to primary realizations of the kilogram
definition that have participated in an initial pilot study [4.10] and/or in BIPM.M-K1 through an
algorithm defined by the CCM. The BIPM 6 disseminates the unit of mass from the average mass
of the ensemble. NMIs, DIs, the BIPM or collaborations among them, may adopt a similar
strategy for dissemination of the mass unit.
Preserving the continuity of measurements traceable to an SI unit before and after its redefinition
is a generally accepted criterion for revised definitions of SI base units. The previous definition
of the kilogram was based on the mass of the international prototype of the kilogram (IPK)
immediately after the prescribed cleaning procedure. The dissemination of the mass unit therefore
required traceability to the mass of the IPK.
A pilot study was performed in 2016 to prepare for the redefinition of the kilogram [4.10]. The
comparison included all available experiments capable of determining the value of the Planck
constant to high accuracy.
In preparation for the redefinition of the kilogram (and other units) the Committee on Data for
Science and Technology (CODATA) Task Group on Fundamental Constants evaluated all
published experimental values for the Planck constant h by July 1st 2017 and recommended the
numerical value of h to be used for the new definition of the kilogram [5.3]. The relative
uncertainty of h recommended by the Task Group was assigned to the international prototype of
the kilogram just after fixing the numerical value of h. As a consequence the 26th CGPM
confirmed in its Resolution 1 that, just after the redefinition, the mass of the IPK was still 1 kg,
but within an uncertainty of 1.0 × 10–8. Accordingly, all mass values traceable to the IPK were
unchanged when the new definition came into effect, but all associated uncertainties of these
6
The BIPM operates under a quality management system (QMS) that conforms to ISO/IEC 17025:2005.
The QMS is under the exclusive supervision of the CIPM. Competence is demonstrated through on-site
audits conducted by external experts and regular reports to CIPM Consultative Committees and Regional
Metrology Organizations.
6/14
SI Brochure – 9th edition (2019) – Appendix 2 20 May 2019
mass values were increased by a common component of relative uncertainty, equal to the relative
uncertainty of the IPK just after the redefinition.
Subsequent changes to the mass of the IPK may have historical interest even though the IPK no
longer retains a special status or a dedicated role in this mise en pratique [5.4]. By following the
change in mass of the IPK over time, one may be able to ascertain its mass stability with respect
to fundamental constants, which has long been a topic of conjecture. For that reason, the IPK and
its six official copies are conserved at the BIPM under the same conditions as they were prior to
the redefinition.
References
[2.1] BIPM, The International System of Units (SI Brochure) [9th edition, 2019],
https://www.bipm.org/en/publications/si-brochure/.
[2.2] Richard P, Fang H and Davis R, “Foundation for the redefinition of the kilogram”,
Metrologia 53 (2016) A6–A11.
[3.1] OIML D28 (2004), https://www.oiml.org/en/files/pdf_d/d028-e04.pdf
[3.2] Robinson I A and Schlamminger S, “The watt or Kibble balance: a technique for
implementing the new SI definition of the unit of mass”, Metrologia 53 (2016) A46–
A74.
[3.3] Shaw G A, Stirling J, Kramar J A, Moses A, Abbott P, Steiner R, Koffman A, Pratt J R
and Kubarych Z J, “Milligram mass metrology using an electrostatic force balance”,
Metrologia 53 (2016) A86–A94.
[3.4] Davis R S and Milton M J T, “The assumption of the conservation of mass and its
implications for present and future definitions of the kilogram and the mole”, Metrologia
51 (2014) 169–173.
[3.5] Cladé P, Biraben F, Julien L, Nez F and Guellati-Khelifa S, “Precise determination of the
ratio h/mu: a way to link microscopic mass to the new kilogram”, Metrologia 53 (2016)
A75–A82.
[3.6] Fujii K, Bettin H, Becker P, Massa E, Rienitz O, Pramann A, Nicolaus A, Kuramoto N,
Busch I and Borys M, “Realization of the kilogram by the XRCD method”, Metrologia
53 (2016) A19-A45.
[4.1] CCM Recommendation G 1 (2017),
https://www.bipm.org/cc/CCM/Allowed/16/06E_Final_CCM-Recommendation_G1-
2017.pdf
[4.2] Stock M, Davidson S, Fang H, Milton M, de Mirandés E, Richard P and Sutton C,
“Maintaining and disseminating the kilogram following its redefinition”, Metrologia 54
(2017) S99-S107.
[4.3] Paper on the BIPM ensemble of reference mass standards
[4.4] Nielsen L, “Disseminating the unit of mass from multiple primary realisations”,
Metrologia 53 (2016) 1306-1316.
[4.5] Measurement comparisons in the CIPM MRA, CIPM MRA-D-05, Version 1.6 (March
2016), http://www.bipm.org/utils/common/CIPM_MRA/CIPM_MRA-D-05.pdf
[4.6] Link to BIPM.M-K1
[4.7] Traceability in the CIPM MRA, CIPM 2009-24 (revised 13 October 2009)
https://www.bipm.org/utils/common/documents/CIPM-MRA/CIPM-MRA-
Traceability.pdf
[4.9] Resolution 1 of the 25th CGPM (2014),
http://www.bipm.org/utils/common/pdf/CGPM-2014/25th-CGPM-Resolutions.pdf
7/14
SI Brochure – 9th edition (2019) – Appendix 2 20 May 2019
8/14
SI Brochure – 9th edition (2019) – Appendix 2 20 May 2019
ANNEXES
9/14
SI Brochure – 9th edition (2019) – Appendix 2 20 May 2019
The unified atomic mass constant mu is defined in terms of the mass of the 12C isotope
mu = m(12C)/12 . (A1.1)
The unified atomic mass unit, u, also known as the dalton (symbol: Da), is not an SI unit.
Formally, the conversion between u and kg is u = {mu} kg where the curly brackets around mu
mean “the numerical value of mu when it is expressed in the unit kg”.
where Ar(X) is the relative atomic mass of X, and m(X) is the atomic mass of X. (Relative atomic
mass is usually called “atomic weight” in the field of chemistry.) The elementary entity X must
be specified in each case. If X represents an atomic species, or nuclide, then the notation AX is
used for a neutral atom where A is the number of nucleons; for example: 12C.
In the SI, mu is determined experimentally in terms of the definition of the kilogram. See the next
section for additional information.
The molar mass of X, M(X), is defined as the atomic mass of the entity X multiplied by the
Avogadro constant, NA. The SI coherent unit of M(X) is kg mol-1. For any elementary entity X,
M(X) is related to m(X) through NA:
Mu = M(12C)/12 . (A1.4)
These four equations relate the various quantities which are the building blocks of atomic and
molar masses and, by extension, are often applied to subatomic and molecular masses.
1
hR∞ = meα 2 c , (A1.5)
2
where R∞ is the Rydberg constant, me ≡ m(e) is the electron rest mass, α is the fine-structure
constant and c is the speed of light in vacuum.
First, it follows from (A1.2) and (A1.5) that for any entity X,
h 1 Ar ( e ) α 2 c
= . (A1.6)
m ( X ) 2 Ar ( X ) R∞
10/14
SI Brochure – 9th edition (2019) – Appendix 2 20 May 2019
NAh 1 α 2c
= Ar ( e ) . (A1.7)
Mu 2 R∞
The right-hand side of (A1.7), which is traceable to the SI units of time and length, has a relative
standard uncertainty of 4.5 × 10–10 [5.3] at the time of the revision of the SI. This relation is key
to understanding how the uncertainties of Mu and mu were affected by Resolution 1 of the 26th
CGPM (2018).
Of the constants appearing in the seven relations shown above, Mu (and by extension M(12C)),
had a fixed numerical value before the SI was revised by the 26th meeting of the CGPM, but no
longer. The constants NA and h did not have fixed numerical values prior to the 26th CGPM. (The
value of the speed of light in vacuum has been fixed since 1983).
Thus Resolution 1 of the 26th CGPM has had the following consequences to the quantities and
measurements discussed above:
1. Relative atomic masses (and their uncertainties) are unaffected. They are dimensionless ratios
and thus independent of unit systems. In the field of chemistry, relative atomic masses are often
referred to as atomic weights.
3a. Neither the value nor the uncertainty of NAh/Mu were affected by Resolution 1. The value of
this combination of constants is still determined from the recommended values for the parameters
on the right-hand side of (A1.7), and these are either traceable to SI units of time and length or
are pure numbers.
In some scientific papers published prior to the adoption of Resolution 1, the quantity NAh/Mu has
been written as NAh(103), where the factor 103 was used as a kind of short-hand to indicate the
exact numerical value of Mu-1 whose SI coherent unit is mol kg-1. This short-hand arose because
the mole was defined through the definition of the kilogram combined with an exact numerical
value of Mu equal to 10–3 kg mol–1; but the mole is now defined through a fixed numerical value
of NA, whose SI coherent unit is mol-1. Nevertheless, Mu may still be taken to be 0.001 kg mol-1 as
long as the relative standard uncertainty of Mu, which is currently 4.5 × 10-10 [5.3], can be
neglected in the uncertainty budget of a measurement under discussion.
3b. For no other reason than to bring clarity to the discussion in this subsection, the changes to
the value of Mu and its uncertainty may be parameterized in terms of a small, dimensionless
quantity k. The molar mass constant Mu, instead of being defined as exactly 0.001 kg mol-1, as it
was prior to the adoption of Resolution 1, can be accurately derived from the last term of the
following relation
R∞ 2 N A h
=Mu mol ) (1 + k )
( 0.001 kg= −1
,
Ar ( e ) α 2 c
(A1.8)
where, in the last term, the constants in the final parentheses have exactly defined values.
Due to the principle of continuity when changes are made to the SI, the value of k is consistent
with zero to a standard uncertainty of u(k) = ur(R∞/(Ar(e)α2)), which at present is 4.5 parts in 1010.
This uncertainty would be further reduced by improved measurements of the constants involved,
11/14
SI Brochure – 9th edition (2019) – Appendix 2 20 May 2019
α in particular. The accepted values and relative uncertainties of Ar(e), R∞ and α are the
CODATA 2017 recommended values [5.3].
The molar mass constant and the unified atomic mass constant are related by Mu = muNA. It
follows that, since ur(NA) = 0, the relative uncertainties of mu and Mu are identical:
For the case of mu, whose value has been (and remains) determined by experiment, the adoption
of Resolution 1 nevertheless resulted in a reduction of ur(mu) by more than a factor of 20 simply
by defining h to have a fixed numerical value, although this improved uncertainty does not seem
to have any immediate practical benefits.
Finally, in atomic physics it is sometimes necessary to convert between the non-SI units
electronvolt (symbol: eV) and the unified atomic mass unit (symbol: u). The correspondence is at
present
where the numerical value of the energy expressed in electronvolts equals the numerical value of
muc2/e expressed in joules per coulomb. The quantities c and e have fixed numerical values.
In the past, an experiment capable of determining the value of the Planck constant provided a
result of enduring value, even if the experiment was never repeated. Now that similar
experiments are used to realize the mass unit, we discuss briefly whether an abbreviated
experiment could be used to ensure that the realization remains valid. If we consider the
realizations described in Section 2, the basic question is: must routine realizations of primary
mass standards be identical to the first such realization? Some considerations are given here.
For realization through a Kibble balance: Assurances are needed that the mechanical and
magnetic alignments of the balance remain adequate; that SI traceability is maintained to
auxiliary measurements of velocity, gravitational acceleration, current and voltage. Improved
technology in these areas opens the possibility of reducing the uncertainty of the realization.
For a realization through the XRCD method, 28Si-enriched, single-crystal silicon ingots were
prepared. X-ray interferometers, samples for molar mass measurements, two 1 kg spheres for the
density measurement, and many other samples were prepared from each ingot. The spheres are
primary mass standards from which the mass unit can be disseminated, but the spheres must be
maintained in good condition for periodic monitoring by appropriate methods of the following
parameters:
• Surface layers on the silicon spheres by, for example, spectral ellipsometry, X-ray
refractometry (XRR), X-ray photoelectron spectrometry (XPS), X-ray fluorescence
(XRF) analysis, and infrared absorption;
• Volume of the silicon spheres by, for example, optical interferometry.
These measurements are not onerous and it is estimated that they could be carried out within a
few weeks.
12/14
SI Brochure – 9th edition (2019) – Appendix 2 20 May 2019
In addition, although no known mechanism would change the molar mass of the crystals,
re-measurement of the molar mass by improved methods could reduce the uncertainty with which
the kilogram definition can be realized by the XRCD method.
Similarly, there is no known mechanism for the edge dimension a(Si) of the unit cell to change
with respect to time, but re-measurement of this quantity by combined X-ray and optical
interferometry could reduce the uncertainty with which the kilogram definition can be realized by
the XRCD method.
Confirmation can be provided by mechanisms of the CIPM MRA, which provide measures of the
equivalence of the various realizations.
In the context of the CIPM MRA, an NMI, DI or the BIPM, realizing the mass unit would be able
to calibrate mass standards traceable to their own realization only, provided that the laboratory
has participated with success in a key comparison as described in section 3.1. However, as long
as the uncertainty of a primary realization is significantly larger than the uncertainty of a mass
comparison, the uncertainty of a calibration traceable to a single realization would be larger than
the uncertainty of a calibration traceable to multiple realizations at least in the case of
independent and consistent results.
Laboratories realizing the mass unit might take advantage of the information obtained in key
comparisons in order to reduce the mass calibration uncertainty and increase the correlation of
mass measurement worldwide. The following simplified example illustrates how the analysis of
the key comparison might be modified in order to achieve this.
Assume that a number n of laboratories is realizing the mass unit. These laboratories are labeled
NMI1,…,NMIn. As a result of the realization, NMIi assigns a prior value mi and an associated
standard uncertainty u(mi) to a stable mass standard Si with nominal mass 1 kg. In a subsequent
key comparison, NMIi measures the mass difference between the standard Si and a circulated,
stable mass standard SR. NMIi reports the measured mass difference ∆mi, the prior mass value mi
and the associated standard uncertainties u(∆mi) and u(mi).
The key comparison reference value m̂R (the mass of the circulated standard SR) and highly
correlated posterior values m̂i of the mass standards Si are obtained as the weighted least squares
solution to the model
𝑚𝑚1 1 0 … 0 0
𝑚𝑚2 0 1 … 0 0
⎛ ⋮ ⎞ ⎛ 𝑚𝑚
�1
⋮ ⋮ ⋱ ⋮ ⋮⎞
⎜ 𝑚𝑚 ⎟ ⎜ ⎟ 𝑚𝑚
�
⎜ 𝑛𝑛
⎟ =̇ ⎜ 0 0 … 1 0 ⎟ ⎛ ⋮2 ⎞
(A3.1)
⎜ ∆𝑚𝑚1 ⎟ ⎜1 0 … 0 − 1⎟ ⎜𝑚𝑚
� 𝑛𝑛
⎟
⎜∆𝑚𝑚2 ⎟ ⎜0 1 … 0 −1 ⎟
𝑚𝑚
�
⋮ ⋮ ⋮ ⋱ ⋮ ⋮ ⎝ 𝑅𝑅 ⎠
⎝∆𝑚𝑚𝑛𝑛 ⎠ ⎝0 0 … 1 − 1⎠
(The symbol =̇ , also used in [5.3], indicates that an input datum of the type on the left-hand side
is ideally given by the expression on the right-hand side containing adjusted quantities.)
13/14
SI Brochure – 9th edition (2019) – Appendix 2 20 May 2019
In the subsequent dissemination of mass unit, NMIi uses the stable mass standard Si as reference,
but with the posterior value mˆ i and associated standard uncertainty u( mˆ i ) rather than the prior
value mi and associated standard uncertainty u(mi).
For simplicity, the above example is based on the assumption that stable mass standards are
available. Such standards were not available in the past, and they may not be available in the
future either. However, as long as the changes in mass standards are predictable with an
uncertainty smaller than the uncertainty of the realization of the mass unit, a procedure similar to
the one described, but which takes into account the instability of the mass standards, will provide
posterior mass values with smaller uncertainties and higher correlations than those of the prior
values.
14/14